freight demand and economic activities: modelling ......freight demand and economic activities:...

Freight demand and economic activities: modelling connections and multimodal

transport networks

prof. Ennio Cascetta

Andrea Papola, Vittorio Marzano, Fulvio Simonelli

The port and maritime sector: key developments and challenges

Antwerp, May 11-12, 2015

Outline

– introduction

– freight supply models

– freight demand models

– towards NEG models

– the container revolution and freight modelling challenges

Freight demand and supply chains

Freight demand is the result of complex supply chains, involving

all geographical levels …

Inte

rmedia

te g

oods

raw

mate

rials

Freight modeling

• two main types of models:

– private oriented, related to optimization and planning/management of company-related operations)

– public oriented, for system-wide and social planning and governance

• why mathematical models for planning and governance?

– forecasts and DSS

– missing data in current scenario:

• unavailability

• unsatisfactory coverage

• unfeasible economic and time collection efforts

• this presentation:

– international and national models for planning and governance

Freight modeling

Planning and governance applications:

– design and assessment of transportation policies:

• infrastructures (e.g. new road/rail links, new ports or logistics platforms)

• regulation (e.g. time windows constraints for pickup/delivery, weight limits for lorries)

• fiscal (e.g. gasoline taxes, subsidies to intermodal transport)

– assessment of economic impacts of transportation policies:

• variations in industrial production

• GDP impacts

• number of jobs

• spending effects

• …

Freight supply models

• planning and management of freight systems requires implementation of multimodal freight supply models

• a challenging task, especially in large geographical areas, because of three issues not properly addressed by existing literature:

– presence of non-additive transport costs, e.g. EU regulation on rest/stop times for road drivers or non-additive fares/tariffs for rail and sea modes

– multiple freight segments (e.g. for road/sea Ro-Ro transport: 1 driver/2 drivers, own account/hiring, accompanied/unaccompanied and so on)

– not straightforward multimodal integration of single modes

Euro-Mediterranean model:

1508 traffic zones (NUTS3)

Study area and zoning

Network models: example

ROAD

SEA IWW

RAIL

Network models: example

mode # links km links # nodes

road 63.109 704.988 50.870

sea 28.319 n.s. 2.147

inland waterways 1.041 22.347 966

rail 90.259 406.973 83.462

The problem of non-additive impedances

• non-additive transport costs:

– nonlinear fares/tariffs

– EC 561/2006 on rest/stop times for drivers, simplifying:

• daily driving period shall not exceed 9 hours, with an exemption of twice a week when it can be extended to 10 hours

– superadditivity holds for monomodal networks (time and/or time-proportional cost impedances):

– … but it does not hold for multimodal networks!

)()( dBsBdAsAdBdA tttttt

)( )( dBsBdBtotBdAsAdAtotA tttttttt

Super-additivity on multimodal graphs


The shortest path on

multimodal network

accounting only for additive

link travel times is all-road

(duration=20 h) …

travel times in hours


… but it leads to additional 16 hours of

rest time (1 driver) or 8 hours (2 drivers),

for a total of 36 hours.

Therefore, the overall shortest path is the

following (22 hours) …

travel times in hours

macroarchi stradali

archi di trasferimento

modale

macroarchi marittimi

centroidi

terminali

terminali clonati


• superadditivity does not hold on multimodal networks:

– need for “virtual” networks with macro-links representing overall connections between centroids and intermodal terminals

• need for calculating LOS related to specific mode sequences:

road macrolinks

mode transfer links

sea macrolinks

centroids terminals “cloned” terminals

ROAD ROAD SEA

Cen

tro

ids

Po

rts

(lan

dsi

de)

Po

rts

(sea

sid

e)

Clo

ned

po

rts

(sea

sid

e)

Clo

ned

po

rts

(lan

dsi

de)

Cen

tro

ids

Analytical cost model (cost functions)

• several supply segments, e.g. for road/sea Ro-Ro transport:

– 1 driver/2 drivers

– own account/hiring

– accompanied/unaccompanied (i.e. tractor+trailer or only trailer)

– …

Analytical models: example

Sevilla Valencia

Barcelona

Milan

Genoa

lower VTTS for

trailers alone

leads to longer

maritime routes

Analytical models: example

lower VTTS for

trailers alone

leads to longer

maritime routes

Milan

Trieste

Ancona

Bari

Bar

Durres

Tirana

lower VTTS for

unaccompanied

leads to longer

maritime routes

the 1 driver/2

drivers option

changes the

maritime route

Road network

Sea truck+trailer (1 dr)

Sea only trailer

Sea truck+trailer (2 dr)

All road

Freight demand models

Freight demand simulation models may follow in principle the

traditional “four step” structure of the passenger demand models:

EMISSION/

ATTRACTION

DISTRIBUTION

Econometric models

Regression models

Econometric models

Growth factor models

Regression models

MODE

ROUTE

Traditional mode choice models

Mode choice models with logistic attributes

Logistic models freight transport service choice rather than mode choice

Gravity

models

economic

activity

models

Path choice models

Gravity models

singly constrained

Go = f(Production unitso, Production levelo)

Ad = f(Consumerd, Consumption leveld)

f(cod) = function inversely proportional to costod e.g. exp[-cod]

in origin in destination

supply driven demand driven

MNL distribution

percentages MNL acquisition

percentages

'

'' )(

)(

d

odd

oddood

cfA

cfAGD

'

'' )(

)(

o

doo

ododod

cfG

cfGAD

Gravity models

doubly constrained

Dod = kokdGoAdf(cod)

f(cod) = function inversely proportional to costod

Sd Dod = Go

So Dod = Ad

ko = 1/Sd kdAdf(cod)

kd = 1/So koGof(cod)

Constants can be calculated in a double constrained gravity model through

an iterative algorithm (given a cost function)

explicit representation of interdependence among different

sectors of economy

the economic structure of production by sector, for a given

region, can be represented by its input-output table:

MRIO models

production usage (output)

production mix (input)

S1 S2 S3 S4

S1 … … … … … … …

S2 … … … … … … …

S3 … K 32i … … Y 2

i Y REG2

i Y EST2

i

S4 … … … … … … …

Added value … … … …

… X 2i … …

Regional import … J REG2

i … …

International import … J EST2

i … …

Regional

export

International

export

Sectors of production

Value of production

Sectors of productionRegion i Final demand

MRIO models

Ym

Xm Xm

Xm

Xm

The consequences of sectorial interdependency:

A

B E

C D

X= production

Y= consumption

m= a final product (e.g. cars)

MRIO models

Xn

Xn

Xn

Xn

Xm


A

B E

C D

X= production

Y= consumption

m= a final product (e.g. cars) – n= an intermediate input (e.g. engines)

MRIO models

Xk

Xk

Xk

Xk

Xn


A

B E

C D

X= production

Y= consumption

n= an intermediate input (e.g. engines) – k= a raw material (e.g. steel)

MRIO models

Ym

Xm Xm

Xm

Xm

Xn

Xn

Xn

Xn

Xm Xk

Xk

Xk

Xk

Xn


A

B E

C D

DYm

DXn, DXk, DXj, …

DXm X= production

Y= consumption

MRIO models

Solving for X, the functional form of the MRIO model is:

and therefore the data needed for estimation/application are:

– vectors of:

• international exports (YEXT)

• final demand (Y)

– matrices of:

• technical coefficients (A)

• trade coefficients (T)

X=(I-TA)-1[T(Y+YEXT)]

X =TAX + T(Y+YEXT) X(I-TA) = [T(Y+YEXT)]

MRIO model and transport modelling

trade coefficient tmij percentage of sector m products acquired

from zone i and used in zone j for whatever use.

trade coefficients can be modelled as a function of level of service attributes, representing generalized transport cost between zones

explicit representation of the influence of transport system on freight production and distribution and therefore on the economic structure of each region

The RUBMRIO Model

trade coefficients can be modeled explicitly as a function of:

• transportation level of service attributes

• selling prices

• …

acquisition zone i is a compound alternative composed of the aggregation of a number EDm

i of elementary origins for sector m

a Nested Logit model may be used to model tmij

various possible feedbacks:

• short term: trade coefficients elastic to transport LOS only

• long term: also significant macroeconomic feedbacks

k

mm

kj

mm

ijm

ijV

Vt

)/exp(

)/exp(

k

mm

k

mm

kj

mm

i

mm

ijm

ijsfVexp

sfVexpt

]/)/[(

]/)/[(

The RUBMRIO model: short term

MRIO model

final demand by sector and country

trade coefficients

model

average transport costs for all modes

other attributes

freight supply model

freight matrix in value

value/quantity transformation

base reference matrix in quantity

freight matrix in quantity

GDP and macroeconomic

variables

mode choice model

freight matrix in quantity by mode

technical coefficients

trade coefficients

in value

LOS

selling prices

MR

IO M

OD

EL

T

RA

NS

PO

RT

M

OD

EL

The RUBMRIO model: long term

MRIO model

final demand by sector and country

trade coefficients

model

other attributes

freight supply model

freight matrix in value

value/quantity transformation

base reference matrix in quantity

freight matrix in quantity

GDP and macroeconomic

variables

mode choice model

freight matrix in quantity by mode

selling prices model

technical coefficients

trade coefficients

in value

LOS

selling prices

MR

IO M

OD

EL

T

RA

NS

PO

RT

M

OD

EL

average transport costs for all modes

Difficult expectations!

Macroeconomic impacts generated by change in transport costs are very difficult to predict:

• countries with economic competitive advantages may benefit more from a decrease of transport costs (and vice versa), being able to increase their export

• competitive advantages may depend on different factors (availability of raw materials, developed technologies, low labor cost etc.) which are implicitly accounted for in the technical coefficients

• a tout court decrease of transport costs impacts differently on Countries depending on their geographical position within the study area, i.e. of their initial accessibility with respect to the other Countries

MRIO model: modelled feedbacks

Example with two zones (A,B) and only one traded product

impact of transport changes in a MRIO model

Towards the NEG paradygm

• job market and migration

• cross-level of salaries

• final demand

• production costs of

intermediate and final inputs

• …

• currency exchange

• availability of raw materials

• …

http://www.vmapas.com/Immagine_Satellitare_Mondo_Occidentale.jpg/maps-it.html?map_viewMap=1


Input Output

(Leontief)

New Economic Geography (Krugman)

Freight demand and supply chains

… and the “container revolution” made the world smaller and

has allowed worldwide supply chains:

Examples of supply chains

Impacts of containerization

REDUCTION OF VESSEL

LOADING/UNLOADING TIMES

INTENSIVE USE

OF

TRANSSHIPMENT

REGULAR LINER

SERVICES

CONTAINER

NETWORK OF

REGULAR LINER

SERVICES

HIGHER ACCESSIBILITY

FOR LOW-DEMAND O-D

PAIRS

ORGANIZATIONAL

AND TECNOLOGICAL

INNOVATIONS

SCALE AND SCOPE

ECONOMIES

CONCENTRATION

OF FREIGHT FLOWS

REGULAR NETWORK OF

SERVICES FOR GENERAL

CARGO FLOWS

REDUCTION OF TOTAL

TRANSPORT

TIMES/COSTS

Analysis of costs on different trade lanes

0

1.000

2.000

3.000

4.000

5.000

6.000

7.000

8.000

0 2.000 4.000 6.000 8.000 10.000 12.000 14.000 Co

sto

tra

sfe

rim

en

to d

i un

co

nta

ine

r d

a 4

0' (

€)

Capacità del naviglio (TEU)

Economie di scala generate da navi portacontainer sulle tre principali rotte continentali Est-Ovest

Rotta Europa-Far Eastlunghezza pari a ca. 18.500 km

Rotta Trans-Pacif icalunghezza pari a ca. 12.900 km

Rotta Trans-Atlanticalunghezza pari a ca. 6.450 km

Fonte: elaborazioni LOGICA su dati di K. Cullinane, M. Khanna, 2000; W. Coyle, W. Hall, N. Ballenger, 2001; interviste a Compagnie dei Lavoratori Portuali italiane, 2010

Variazione costi: -58%



Scale economies on the three main trade lanes

co

st

for

tran

sp

ort

ing

a 4

0’ co

nta

iner

[€]

vessel capacity [TEU]

Summary

MODEL Data availability

and ease of implementation

Applications

Detailed O-D freight flows

forecasts

effects on GDP and supply

chains

Complex long term economic

feedbacks

Gravity models

MRIO models

RUBMRIO models

NEG models

Thank you!

The port and maritime sector: key developments and challenges

Antwerp, May 11-12, 2015

References

• De la Barra, T. (1989). Integrated Land Use and Transport Modelling. Cambridge University Press.

• Cascetta, E., Di Gangi, M. and G. Conigliaro (1996). A multi-regional input-output model with elastic trade coefficients for the simulation of freight transport demand in Italy. Proceedings of the 24th PTRC.

• Cascetta E. (2001). Transportation Systems Engineering: theory and methods. Kluwer ed.

• Cascetta E., Marzano V. and A. Papola (2008). Multi regional input-output models for freight demand simulation at a national level. In: Recent Developments in Transport Modelling: lessons for the freight sector (Van der Voorde, E. and Meersman, H. eds.), Emerald.

• Jin, L., Kockelman, K. and Y. Zhao (2003). Tracking Land Use, Transport, and Industrial Production using Random-Utility based Multizonal Input-Output Models: Applications for Texas Trade. Proceedings of the 82nd TRB Meeting, Washington.

• Marzano V. and A. Papola (2004). Modelling freight demand at a national level: theoretical development and application to Italian demand. Proceedings of 2004 ETC Conference, Strasbourg.

• Zhao, Y. and K.M. Kockelman (2004). The random utility based multiregional inptut-output model: solution existence and uniqueness. Transportation Research, 38B, 789-807.

RUBMRIO model: example (EU level)

Study area:

– 21 EU Countries

– 2005 base year for model estimation

Type of data source:

input-output tables (for X, YEXT, Y, A)

import/export trade flows in value (for T)

Two examples of application of the proposed system of models:

• Scenario 1: 10% increase in road costs, e.g. coming either from an increase of oil price or from a reduction of public subsidies for lorries

• Scenario 2: 20% reduction of maritime fares, so as to mimic a policy of public subsidies towards motorways of the sea within the European Union

RUBMRIO model: example (EU level)

Applications: results

tons/km2/year generated by NUTS3 zone in the base scenario

Applications: results

tons/km2/year generated by NUTS3 zone in the scenario 2

generated tons

freight demand and economic activities: modelling ......freight demand and economic activities:...

Documents